Assumptions & Risk Map

Most plans hide their beliefs inside their numbers. This page pulls them out, ranks them by how much the strategy depends on them and how little you know, and turns the top ones into small tests with a result that would change your mind.

Practitioner craft with early experimental support. Discovery-driven planning (McGrath & MacMillan, 1995) and assumption-based planning (Dewar, 2002) are untested tools; a randomized trial of startups (Camuffo et al., 2020) supports hypothesis-led testing in general, not these tools. The likelihood and impact grid is openly criticized (Cox, 2008): use it to sort attention, not to compute risk. Scores are rules of thumb.

Use it when:

• You have chosen a direction in 7.8 Preferred Strategy Sheet and are about to commit money, people or your reputation.
• A plan rests on things you have not seen with your own eyes: a partner's pace, customers' willingness to pay, a regulator's reading.
• Conditions have changed since you last looked, or a quarter has passed.

Skip it when:

• You have no strategy yet: go to 7.7 Strategic Option Cards and 7.8 Preferred Strategy Sheet.
• The strategy is already running and you need to track execution risks: use 8.10 Risks, Issues & Decisions Log.

The idea is a reversal of ordinary planning. Conventional plans treat their numbers as facts and bury their beliefs. McGrath and MacMillan (1995) proposed planning the other way around for anything new: write the assumptions down and spend the early budget testing the ones that could kill you. Dewar (2002) built the same move into a method from defense planning: find the load-bearing assumptions, attach signposts that warn when one is breaking, and prepare the response in advance. Ries (2011) popularized the test-and-pivot loop for startups.

All of this is practitioner craft, with early experimental support and published criticism both emerging. One caution: matrices of likelihood and impact are openly disputed (Cox, 2008), so the Risk Map sorts attention and does not measure. The steps follow the method. Scoring ranks beliefs by criticality and uncertainty, because testing cheap certainties first is how teams feel busy while staying exposed. Killer assumptions force the pivot to be named in advance. And every test needs its success criteria written before it runs, since a test that cannot fail is theater.

Copy the statement from 7.8 and list what must be true for it to work, in your own words. A fact you can already show is not an assumption; leave it out.

Strong answer: "Hotels in the second city will accept next-day service on a route twice as long as today's." Not: "Customers will like us." One assumption, checkable, with a customer behavior in it.

Strategy statement (from 7.8):

Time horizon:

Illustrative example: the business and all numbers are made up. It continues the laundry from 7.6 to 7.8: option C with a local laundry as a bridge.

Top assumptions:

• A1 (criticality 5, uncertainty 4): hotels in the second city accept next-day service on a route twice as long as today's.
• A2 (5, 3): the bridge laundry meets our quality standard for the first contracts.
• A3 (4, 3): the second line is installed within four months.
• A4 (4, 4): the regional chain does not undercut with multi-year contracts (rated low confidence in 7.6).

Killer assumptions: A1 (early sign: calls rarely lead to a trial; kill result: fewer than three of ten hotels accept) and A4 (early sign: a hotel names the chain's price). Pivot if A1 is false: a small satellite depot in the second city, or a return to the do-less option from 7.7.

Risks: line delivery slips (signal: the supplier misses the first date; response: extend the bridge); the chain cuts prices (signal: two hotels quote it; response: offer two-year renewals with a service guarantee).

Tests, 90 days: ten conversations with hotel purchasing managers (success: three accept a two-week trial); one trial hotel served on the real route for two weeks (success: all deliveries next day); a quality audit of the bridge laundry on ten sample batches (success: none below our standard).